- Update from the git skeleton files.
- Fix some newly found issues.

I find this format to be better.

It's set in the CI template, so I don't know why it was ignored. Anyway,
set it here and get it working again.

- Add the code-spell pre-commit hook.
- Correct the spelling errors it found.

So you can insert mnpw between some other system and the plugin.

Not much I can do about them.

Still need to deal with multiple perf data in a single line (not sure if
it's technically correct at all).

The parsing doesn't handle it correctly and fails. Obviously tests will
fail until I fix it.

I'm targeting pretty recent Python versions (3.6+) so why not enjoy the
latest syntax and features.

I think it's OK now that the functionality is there, the usability is
there, it's well(-ish) tested.

The Nagios plugin check object logs on internal dealings. The main
function logs on handling the result.

- Use the parsed plugin output in the notification.
- Log more of the plugin output.

- Works on Python 3.6.
- The upsides of dataclass don't help in this case.
- Is faster and uses less memory (although in this case it would hard to
  even measure the difference).

The tests aren't the greatest and binary is broken but I think the
parsing works.

Let's see if that fixes the issue with Python 3.6 in Centos build image.

It's now in the CI template, so not necessary at all. But will only
happen in the Bats test so everything else should be a little faster.

For Python 3.6, which is what the Centos image has (where I build the
executable using Pyinstaller).

I wouldn't call it perfect or thorough, but it's a little better than
before.

Parse perf data from a plugin's output. Test can be improved, but at
least it doesn't fail (that must count for something).